Total Synthesis of (+)-Calyculin A and (−)-Calyculin B: Cyanotetraene

Oct 30, 1999 - Chiral Action at a Distance: Remote Substituent Effects on the Optical Activity of Calyculins A and B. Rama K. Kondru, David N. Beratan...
0 downloads 13 Views 721KB Size
10478

J. Am. Chem. Soc. 1999, 121, 10478-10486

Total Synthesis of (+)-Calyculin A and (-)-Calyculin B: Cyanotetraene Construction, Asymmetric Synthesis of the C(26-37) Oxazole, Fragment Assembly, and Final Elaboration Amos B. Smith, III,* Gregory K. Friestad, Joseph Barbosa, Emmanuel Bertounesque, James J.-W. Duan, Kenneth G. Hull, Makoto Iwashima, Yuping Qiu, P. Grant Spoors, and Brian A. Salvatore Contribution from the Department of Chemistry, the Laboratory for Research on the Structure of Matter, and the Monell Chemical Senses Center, UniVersity of PennsylVania, Philadelphia, PennsylVania 19104 ReceiVed June 22, 1999

Abstract: A convergent total synthesis leading to (+)-calyculin A and (-)-calyculin B (1 and 2), antipodes of the potent, highly selective and remarkably cell-permeable phosphatase inhibitors calyculins A and B, has been achieved. In the preceding paper we outlined the asymmetric synthesis of the C(9-25) spiroketal dipropionate subunit (+)-BC; herein we describe construction of the C(1-8) cyanotetraene, an asymmetric synthesis of the C(26-37) oxazole, fragment assembly and final elaboration to (+)-1 and (-)-2. Highlights of the synthesis include: application of a one-pot three-component Suzuki reaction for the construction of phosphonate A, a bifunctional triene precursor of the light sensitive C(1-8) cyanotetraene subunit, an asymmetric synthesis of the C(26-32) oxazole (-)-D, exploiting the Silks-Odom 77Se NMR protocol to assess enantiomeric purity, construction of the C(33-37) subtarget (-)-E in a highly stereocontrolled fashion via an acyliminium ion, and a concise, highly efficient sequence for fragment assembly and elaboration to (+)-calyculin A and (-)-calyculin B. The synthesis of (-)-2 also confirms the structure of calyculin B, previously based only on spectral comparison with calyculin A.

The calyculins comprise a family of potent, highly selective serine-threonine phosphatase inhibitors possessing a striking array of stereochemical and functional elements. In this, the second of a two-part full account,1 we describe the completion of a highly convergent total synthesis of (+)-1 and the first total synthesis of (-)-calyculin B (2).2-4 From the outset, we envisioned an approach (Scheme 1) which would provide both 1 and 2 from a common advanced intermediate, avoid extensive manipulations of the light-sensitive5 C(1-9) cyanotetraene, and permit flexibility in fragment coupling. Accordingly, disconnections at the C(2) and C(8) olefins led to phosphonate A (Scheme 1), possessing a latent C(3) carbonyl for penultimate Peterson olefination to access both 1 and 2. Disconnection of BCDE at the C(25-26) olefin revealed subtargets BC and DE. Having described an efficient route to (+)-BC in the preceding (1) Part 1: Smith, A. B., III; Friestad, G. K.; Barbosa, J.; Bertounesque, E.; Hull, K. G.; Iwashima, M.; Qiu, Y.; Salvatore, B. A.; Spoors, P. G.; Duan, J. J.-W. J. Am. Chem. Soc. 1999, 121, XXXX. (2) At the start of our work, the absolute stereochemistry was unknown. (3) Matsunaga, S.; Fusetani, N. Tetrahedron Lett. 1991, 32, 5605. (b) Hamada, Y.; Tanada, Y.; Yokokawa, F.; Shioiri, T. Tetrahedron Lett. 1991, 32, 5983. (4) Preliminary communications: (a) Smith, A. B., III; Friestad, G. K.; Duan, J. J.-W.; Barbosa, J.; Hull, K. G.; Iwashima, M.; Qiu, Y.; Spoors, P. G.; Bertounesque, E.; Salvatore, B. A. J. Org. Chem. 1998, 63, 7596. (b) Smith, A. B., III; Cho, Y. S.; Friestad, G. K. Tetrahedron Lett. 1998, 39, 8765. (c) Iwashima, M.; Kinsho, T.; Smith, A. B., III. Tetrahedron Lett. 1995, 36, 2199. (d) Smith, A. B., III; Iwashima, M. Tetrahedron Lett. 1994, 35, 6051. (e) Salvatore, B. A.; Smith, A. B., III. Tetrahedron Lett. 1994, 35, 1329. (f) Smith, A. B., III; Salvatore, B. A.; Hull, K. G.; Duan, J. J.-W. Tetrahedron Lett. 1991, 32, 4859. (g) Smith, A. B., III; Duan, J. J.-W.; Hull, K. G.; Salvatore, B. A. Tetrahedron Lett. 1991, 32, 4855. (5) Matsunaga, S.; Fujiki, H.; Sakata, D.; Fusetani, N. Tetrahedron 1991, 47, 2999.

Scheme 1

paper, we detail here the synthesis and reactivity of phosphonate A, a concise, highly stereocontrolled route to (+)-DE, the union of subtargets A, (+)-BC, and (+)-DE, and completion of the total synthesis of (+)-calyculin A and (-)-calyculin B.

10.1021/ja992135e CCC: $18.00 © 1999 American Chemical Society Published on Web 10/30/1999

Synthesis of (+)-Calyculin A and (-)-Calyculin B Scheme 2

Evolution of Phosphonate A. Initially we envisioned introduction of the C(1-9) cyanotetraene in one step employing Horner-Emmons reactions of unsaturated phosphonates 3a and 3b (Scheme 2) with a C(9) aldehyde derived from intermediate BCDE. Phosphonates 3a and 3b in turn would arise via Peterson olefination of ketophosphonate 4 with trimethylsilylacetonitrile. To fabricate 4 we planned to exploit a one-pot three-component Suzuki coupling6 of E-bromovinyl boronate 7 with vinyl iodide 6a or 6b and organozinc 8, followed by introduction of the phosphonate and mild acid hydrolysis of the intermediate enol ether. Importantly, the C(8) methyl could be present from the outset (i.e., 6b), or introduced via R-methylation of the corresponding primary phosphonate. We initiated the synthesis of 3a and 3b with known vinyl iodides 6a and 6b, available via Negishi carbometalation (Scheme 3).7 Exposure of E-bromovinyl boronate 7 to vinyl

J. Am. Chem. Soc., Vol. 121, No. 45, 1999 10479 Scheme 4

the three steps was 55%. Unfortunately, R-methylation of 12 could not be achieved; treatment of 12 with a variety of bases led to severe olefin isomerization. This result did not build confidence in our ability to generate metalated phosphonate reagents in the presence of an unsaturated nitrile. We therefore decided to postpone Peterson olefination until after C(8-9) olefin construction.9 At this stage we also learned that introduction of the phosphonate group could not be readily achieved when C(8) was substituted with a methyl group (Scheme 5). For example, Scheme 5

Scheme 3

attempts to convert allylic chloride 14 to phosphonate 4 using the standard Arbuzov8 or Becker10 conditions proved unsuccessful. We therefore modified our approach to include the phosphonate prior to Suzuki coupling. In particular, we chose to employ a delicate balance of reactive functionality at the termini of the C(1-8) synthon (Scheme 6), which would require

zinc 8 in the presence of bis(triphenylphosphine)palladium(II) chloride (5 mol %) provided dienylboronate 9 (not isolated), which upon addition of a lithium hydroxide-methanol solution and vinyl iodide 6a (or 6b) established the C(3-8) carbon skeleton in the form of enol ethers 5a,b; hydrolysis with dilute aqueous hydrochloric acid furnished ketones 10a (or 10b), albeit in variable but good yields. To test the viability of the Horner-Emmons tactic in the presence of the conjugated nitrile (i.e., after Peterson olefination), 10a was treated with trimethylsilylacetonitrile and lithium bis(trimethylsilyl) amide; a mixture of E/Z olefins ( ∼2:1) resulted (Scheme 4), with the major isomer possessing the requisite C(2,3) E geometry as in calyculin B. Removal of the TBS group followed by conversion to the primary mesylate and Arbuzov reaction8 with neat trimethyl phosphite and excess sodium iodide furnished phosphonate 12; the overall yield for

Scheme 6

(6) Ogima, M.; Hyuga, S.; Hara, S.; Suzuki, A. Chem. Lett. 1989, 1959. (b) For a review on Suzuki coupling, see: Miyaura, N.; Suzuki, A. Chem. ReV. 1995, 95, 2457. (7) Rand, C. L.; Van Horn, D. E.; Moore, M. W.; Negishi, E.-I. J. Org. Chem. 1981, 46, 4093. Zirconium-catalyzed carboalumination, followed by quenching with iodine and protection of the allylic hydroxyl as the t-butyldimethylsilyl ether furnished 6a,b from the corresponding propargylic alcohols. (8) Bhattacharya, A. K.; Thyagarajan, G. Chem. ReV. 1981, 81, 415.

minimal manipulations after the Suzuki protocol. In the end, the successful strategy comprised direct use of the delicate enol ether derived from the Suzuki coupling (i.e., A), containing the phosphonate moiety for subsequent C(8) methylation and C(89) Horner-Emmons coupling. Peterson olefination would then be performed after generation of the C(8,9) olefin. Inclusion of the requisite phosphonate on the vinyl iodide (16) would also minimize the number of transformations performed on the sensitive trienol ether. (9) The difficult methylation was consistent with reports by Evans et al. of the low reactivity of 13, prepared by a different method, which did not undergo normal Horner-Emmons coupling, see: Evans, D. A.; Gage, J. R.; Leighton, J. L. J. Am. Chem. Soc. 1992, 114, 9434. (10) Engel, R. Synthesis of Carbon-Phosphorous Bonds; CRC Press: Boca Raton, FL, 1988; p 7.

10480 J. Am. Chem. Soc., Vol. 121, No. 45, 1999 Scheme 7

Preparation of phosphonate A is illustrated in Scheme 7. Conversion of alcohol 187 to the allylic chloride, followed by Arbuzov reaction with neat trimethyl phosphite and sodium iodide provided primary phosphonate 16 in 70% yield for two steps. This iodide proved quite unstable to the usual Suzuki coupling promoters (e.g., lithium methoxide and thallium hydroxide).6b,11 However silver(I) oxide utilized by Kishi was found to be both mild and efficient in promoting the coupling of 16 with dienylboronate intermediate 19.11 In the event, treatment of vinyl iodide 16 at reflux with 19 (2 equiv, formed in situ) and Ag2O (5-6 equiv) in aqueous THF provided 20 in approximately 60% yield. Although 20 could not be fully purified without significant decomposition, C(8) methylation (n-BuLi, MeI, -78 °C) occurred without complications, thereby completing assembly of phosphonate A.12 With phosphonate A in hand, we undertook a model study to explore both the Horner-Emmons coupling and Peterson olefination. Metalation of A (n-BuLi/THF), followed by introduction of model aldehyde 21, provided a 5:1 (E/Z) mixture of C(8-9) olefins; hydrolysis of the enol ether (silica gel) gave ketone 22 in 74% yield. Peterson olefination then furnished cyanotetraene 23 as a 2:1 E/Z mixture (Scheme 8). These results Scheme 8

suggested that phosphonate A, containing the sensitive bifunctional triene, would serve us well. Evolution of the DE Subunit. Disconnection at the amide linkage of DE revealed two fragments, the C(26-32) oxazole D and lactam E (Scheme 9). Reduction of the azide in oxazole D followed by reaction with the acyl-activated lactam E would provide the DE subunit. (11) Uenishi, J.-I.; Beau, J.-M.; Armstrong, R. W.; Kishi, Y. J. Am. Chem. Soc. 1987, 109, 4756. (12) Unstable phosphonate A was best stored in a frozen benzene solution, from which anhydrous A could be obtained upon concentration in vacuo (with azeotropic removal of adventitious moisture). Under these conditions, storage of A for up to one month had no detrimental effect on subsequent Horner-Emmons reactions.

Smith et al. Scheme 9

Construction of the C(26-32) Oxazole via Davidson Cyclization. Reaction of commercially available 4-chlorobutyryl chloride (24, Scheme 10) with lithiated oxazolidinone 2513a Scheme 10

followed by displacement of the primary chloride with sodium azide furnished (+)-26 in 95% yield for the two steps. Evans asymmetric methylation13b of the sodium enolate of (+)-26 then led to (+)-27 in high yield with excellent diastereoselectivity (94.6% de; HPLC). After separation by flash chromatography, displacement of the chiral auxiliary from (+)-27 (β isomer) with the alkoxide derived from known allylic alcohol 2814 and subsequent ozonolysis of the exomethylene gave keto ester 29. Unfortunately, all attempts at Davidson cyclization15 resulted only in low yields ( ∼7%)16 of oxazole 30. Construction of the C(26-32) Oxazole via Oxazoline 31. This approach would entail removal of the chiral auxiliary from (+)-27 (Scheme 11), coupling with a serine derivative, cyclodehydration, and then oxidation. Hydrolytic removal of the chiral auxiliary from (+)-27 (β isomer), exploiting the Evans proto(13) (a) Evans, D. A.; Bartroli, J.; Shih, T. L. J. Am. Chem. Soc. 1981, 103, 2127. (b) Evans, D. A.; Ennis, M. D.; Mathre, D. J. J. Am. Chem. Soc. 1982, 104, 1737. (14) Tanner, D.; Somfai, P. Tetrahedron 1986, 42, 5985. (b) Eliel, E. L.; Badding, V. G.; Rerick, M. N. J. Am. Chem. Soc. 1962, 84, 2371. (15) Davidson, D.; Weiss, M.; Jelling, M. J. Org. Chem. 1937, 2, 328. (16) The Davidson oxazole cyclization has been employed successfully in a variety of cases in which the product oxazoles are 2,4,5-trisubstituted; low yields are typical for 2,4-disubstituted or monosubstituted oxazoles. This is likely due to reduced regiocontrol in enamine formation (i.e., i vs ii) during the cyclization; the reaction is more successful when the substituent at C(5) is aromatic. See Theilig, G. Chem. Ber. 1953, 86, 96.

Synthesis of (+)-Calyculin A and (-)-Calyculin B

J. Am. Chem. Soc., Vol. 121, No. 45, 1999 10481

col,17 provided carboxylic acid (-)-32. Condensation with L-serine methyl ester was then attempted with dicyclohexylcarbodiimide (DCC) and 4-(dimethylamino)pyridine (DMAP); amide (-)-33 was produced in moderate yield (∼65%). Use of diethylcyanophosphonate18 (DECP) as the coupling reagent proved preferable, furnishing (-)-33 in 87% yield. Conversion of the latter to primary chloride 34 with SOCl2 and treatment with silver triflate induced cyclization to afford oxazoline 31.19 Oxidation with nickel peroxide, following the precedent of Meyers et al.,20 furnished D in 40-50% yield (Scheme 11). Scheme 11 Figure 1.

To our delight, 77Se NMR clearly resolved the diastereomeric resonances (Figure 2). Spectrum (a) depicts the 77Se NMR of 37 derived from a known enantiomeric mixture (1.4:1) of oxazole D. Importantly, the resonances from the two diastereomers display baseline separation (∆δ > 2 ppm!). Parts b and Scheme 12

Unfortunately, the 13C NMR spectrum of 31 revealed a 6:1 diastereomeric mixture. Subsequent HPLC analysis of alcohol (-)-33 (>95%) and chloride 34 indicated that epimerization had occurred during chloride formation, not during amide coupling or cyclization. For example, cyclization of pure (-)34 under the above conditions led to a single oxazoline diastereomer. However, a report on the configurational instability of substituents R to the 2-position in oxazolines21 suggested that additional epimerization might be occurring during oxazoline oxidation. The Enantiomeric Purity of Oxazole D. To define the enantiomeric purity of oxazole D, we explored the attachment of a number of chiral auxiliaries to the C(26) carboxylic acid, to the corresponding C(26) alcohol, and to the C(32) amine. For example, reduction of (-)-D to the corresponding amine (H2, Pd/C) and reaction with the Anderson-Shapiro chiral chlorodioxaphospholane22 furnished phosphoramide 35. Unfortunately, 31P NMR analysis of an authentic diastereomeric mixture revealed only broad lines due to P-H(N) coupling; resolution of the NMR resonances was not feasible. Similarly, it was not possible to resolve the diastereomeric resonances of the Mosher esters derived from (-)-D (i.e., 36) by 1H and 19F NMR (Figure 1). We therefore turned to the Silks-Odom selenium-77 NMR technique23 which called for the preparation of 37 (Scheme 12). (17) Evans, D. A.; Britton, T. C.; Ellman, J. A. Tetrahedron Lett. 1987, 28, 6141. (18) Yamada, S.; Kasai, Y.; Shioiri, T. Tetrahedron Lett. 1973, 14, 1595. (19) For example, see: (a) Hamada, Y.; Shibata, M.; Shioiri, T. Tetrahedron Lett. 1985, 26, 5155. (b) Hamada, Y.; Shibata, M.; Shioiri, T. Tetrahedron Lett. 1985, 26, 6501. (20) Evans, D. L.; Minster, D. K.; Jordis, U.; Hecht, S. M.; Mazzu, A. L., Jr.; Meyers, A. I. J. Org. Chem. 1979, 44, 497. (21) Yonetani, K.; Hirotsu, Y.; Shiba, T. Bull. Chem. Soc. Jpn. 1975, 48, 3302. (22) Anderson, R. C.; Shapiro, M. J. J. Org. Chem. 1984, 49, 1304. (23) Silks, L. A., III; Dunlap, R. B.; Odom, J. D. J. Am. Chem. Soc. 1990, 112, 4979. (b) Silks, L. A., III; Peng, J.; Odom, J. D.; Dunlap, R. B. J. Org. Chem. 1991, 56, 6733.

c of Figure 2 depict the corresponding spectra of 37 derived from D prepared via the synthetic routes outlined in Schemes 10 and 11. The diastereomeric resonances in Figure 2b appear in the same ratio ( ∼6:1) as previously observed for chloride 34. We therefore concluded that epimerization did not occur at the crucial C(30) stereogenic center during the Meyers nickel peroxide oxidation (Scheme 11). Three subsequent observations led to the most efficient calyculin oxazole synthesis reported to date (Scheme 13). First, Scheme 13

the degree of epimerization at C(30) in 34 could be reduced (∼< 5%) by employing the milder conditions of MsCl in pyridine. Second, cyclodehydration at 55 °C with the Burgess reagent,24 as introduced by Wipf et al.,25 proved superior to silver triflate and avoided the use of the intermediate chloride and the expensive silver triflate. Third, the Barrish-Singh CuBr2 oxidation26 proceeded in higher yield. Taken together these conditions afforded (-)-D in 67% yield (two steps), with little or no epimerization.27 (24) Burgess E. M.; Penton, H. R., Jr.; Taylor, E. A.; Williams, W. M. Organic Syntheses; Wiley & Sons: New York, 1988; Collect. Vol. VI, p 788. (b) Burgess, E. M.; Penton, H. R., Jr.; Taylor, E. A. J. Org. Chem. 1973, 38, 26. (25) Wipf, P.; Miller, C. P. Tetrahedron Lett. 1992, 33, 907. (26) Barrish, J. C.; Singh, J.; Spergel, S. H.; Han, W.-C.; Kissick, T. P.; Kronenthal, D. R.; Mueller, R. H. J. Org. Chem. 1993, 58, 4494.

10482 J. Am. Chem. Soc., Vol. 121, No. 45, 1999

Smith et al. Construction of Lactam E. Initial efforts to prepare the C(33-37) acid (43) focused on a variation of the Hamada/ Shioiri synthesis28 of the γ-amino acid moiety found in the gastroprotective substance AI-77-B. Beginning with benzylidene aminal 44,28 a known compound available from L-glutamic acid (Scheme 15), selective hydrogenolysis29 followed by methylation of the derived primary alcohol furnished (+)-46 in 90% yield. However, neither hydrolytic opening of the lactam ring nor removal of the N-benzyl group proved viable (Scheme 15). Scheme 15

Although the N-benzyl group could not be removed from lactam (+)-46, the benzylidene group in 44 was easily removed via transfer hydrogenation to provide (+)-47 (Scheme 16).28 Scheme 16

Figure 2.

To facilitate exploratory work, we prepared two additional oxazoles (Scheme 14). Reduction of the azide in (-)-D provided Scheme 14

amine 40, which proved to be unstable as the free base. Alternatively, ester reduction to alcohol (-)-41, followed by azide reduction furnished amine (-)-42 which could be stored without event. (27) Barrish-Singh oxidation did not compromise the integrity of the C(30) stereocenter, as judged by analysis of the optical rotation of (-)-D produced in this fashion.

Installation of p-methoxybenzyl groups on both nitrogen and oxygen, followed by selective removal of the O-PMB group and O-methylation, furnished (+)-49. Oxidative removal of the N-PMB group from (+)-49 could then be achieved with cerium(IV) ammonium nitrate (CAN) in aqueous acetonitrile30 to furnish lactam (-)-50 in 61% yield, accompanied by the over-oxidized imide (-)-51 (24%). The (28) Hamada, Y.; Kawai, A.; Kohno, Y.; Hara, O.; Shioiri, T. J. Am. Chem. Soc. 1989, 111, 1524. (b) Hamada, Y.; Hara, O.; Kawai, A.; Kohno, Y.; Shioiri, T. Tetrahedron 1991, 47, 8635. (29) Sakura, N.; Ito, K.; Sekiya, M. Chem. Pharm. Bull. 1972, 20, 1156.

Synthesis of (+)-Calyculin A and (-)-Calyculin B

J. Am. Chem. Soc., Vol. 121, No. 45, 1999 10483

Table 1

latter could be readily redirected to the synthetic sequence by hydrolysis with LiOOH; a combined yield of 80% for lactam (-)-50 was obtained. The tert-butyl carbamate derivative (-)-E was then generated from lactam (-)-50 in 91% yield; the t-Boc moiety would serve to activate the lactam for ring opening. Although successful, the synthesis of lactam (+)-49 (11 steps from L-glutamic acid, 8.6% overall yield) proved both lengthy and inefficient. We therefore explored an alternate route (Scheme 17), which was envisioned to take advantage of a stereocontrolled addition of an appropriate one-carbon nucleophile to the convex face of N-acyliminium ion 52 derived from hemiaminal 53. The latter would be prepared from erythronolactone; the elegant indolizidine syntheses of Overman and Heitz31 provided precedent for this tactic. Scheme 17

Toward this end, treatment of commercially available isopropylidene-D-erythronolactone [(-)-54]32 (Scheme 18) with the Scheme 18

The reactivity of 56 was tested with n-butylcopper, vinyltrimethylsilane, and trimethylsilylacetylene;35,36 no reaction occurred! Allyltrimethylsilane however reacted efficiently to furnish exclusively the β-allyl adduct (+)-57 in 86% yield (Table 1; entry a). Encouraged by this success, we explored nucleophiles which could be more readily transformed to the required methoxymethyl moiety. Reaction of 56 with trimethylsilylcyanide promoted by boron trifluoride etherate in various solvents (entries b-d) led to 58 in good yield, but with only modest β-selectivities.37 We next explored enol ethers as the nucleophile. Shono and co-workers demonstrated that enol ethers add intermolecularly to N-acyliminium cations to furnish ketones after hydrolysis.38 We reasoned that a ketone which could enolize only in one direction would serve as an effective hydroxymethyl synthon, assuming the sequence of addition, enolization, ozonolysis, and reduction. In the event, treatment of 56 with the trimethylsilyl (TMS) enol ether of pinacolone (61)39 in the presence of BF3‚ OEt2 furnished ketone (+)-59 as a single diastereomer in 81% yield (entry e). The analogous transformation with acetophenone was equally selective, but less efficient (entry f). Generation of the trimethylsilyl enol ether of 59 (Scheme 19) followed by ozonolysis with reductive workup (NaBH4) led as expected to alcohol (+)-48, identical (1H and 13C NMR) to that prepared previously from L-glutamic acid. Importantly, this sequence proceeded in 54% overall yield for the 3 steps. (35) Ludwig, C.; Wistrand, L.-G. Acta Chem. Scand. 1990, 44, 707. (b) Skrinjar, M.; Wistrand, L.-G. Tetrahedron Lett. 1990, 31, 1775. (36) Additions of Grignard species to sulfonylpyrrolidinones i prepared from 53 [MsCl, Et3N; PhSH; monoperoxyphthalic acid magnesium salt hydrate (MMPP); 67% overall] were similarly unsuccessful.

aluminum amide33 derived from p-methoxybenzylamine, followed by Parikh-Doering oxidation and acetylation led to 56 as a 7:1 anomeric mixture in 53% for the three steps.34 (30) Yoshimura, J.; Yamaura, M.; Suzuki, T.; Hashimoto, H. Chem. Lett. 1983, 1001. (31) Heitz, M.-P.; Overman, L. E. J. Org. Chem. 1989, 54, 2591. (32) Lactone (-)-54 can be prepared from D-isoascorbic acid, see: Cohen, N.; Banner, B. L.; Laurenzano, A. J.; Carozza, L. Org. Synth. 1985, 63, 127. (33) Basha, A.; Lipton, M.; Weinreb, S. M. Tetrahedron Lett. 1977, 18, 4171. (34) Single-crystal X-ray analysis of 56 (major anomer) identified it as the β-anomer as expected on steric grounds.

For related examples, see: Brown, D. S.; Hansson, T.; Ley, S. V. Synlett 1990, 48. Brown, D. S.; Charreau, P.; Ley, S. V. Synlett 1990, 749. (37) The β orientation of nitrile 58 (major isomer) was assigned on the basis of 1H NMR coupling constant analysis and confirmed by single-crystal X-ray analysis. (38) Shono, T.; Matsumura, Y.; Tsubata, K. J. Am. Chem. Soc. 1981, 103, 1172. (b) Shono, T.; Tsubata, K.; Okinaga, N. J. Org. Chem. 1984, 49, 1056. (39) House, H. O.; Czuba, L. J.; Gall, M.; Olmstead, H. D. J. Org. Chem. 1969, 34, 2324.

10484 J. Am. Chem. Soc., Vol. 121, No. 45, 1999 Scheme 19

Smith et al. functionality at C(26) also appeared prudent until after the C(34,35) interchange. Synthesis of the second generation C(26-37) subtarget (Scheme 21) began with the union of amino ester 40 with (-)Scheme 21

Union of Fragments C(26-32) and C(33-37). Two methods were investigated to couple the C(26-32) oxazole 30 with the C(33-37) lactam (-)-E (Scheme 20). Reaction of the Scheme 20

aluminum amide derived from amino oxazole 63 (prepared by Lindlar reduction40 of azide 30) with (-)-E was explored initially. This reaction proved extremely slow, furnishing consistent yields of 64 only when a large excess of the aluminum amide (>1.5 equiv) was employed. Better results were obtained by coupling γ-amino oxazole (-)-42 with γ-amino acid (-)65, available by lithium hydroxide hydrolysis of (-)-E;41 diethylcyanophosphonate (DECP) proved to be the coupling agent of choice. With concentrations of (-)-65 and (-)-42 near 0.3-0.4 M, amide (+)-66 was routinely obtained in ∼70% yield. With (+)-66 in hand, formation of primary iodide (+)67, progenitor of the phosphonium salt, was readily accomplished with triphenylphosphine, imidazole, and iodine. A Second Generation C(26-37) Subtarget. In the end, we found that the acetonide at C(34,35) was not a viable protecting group due to its hydrolytic stability in the completely functionalized calyculin framework.42 Although the related cyclopentylidene ketal could be more readily removed, eventual success was achieved with diethylisopropylsilyl (DEIPS) groups43a deployed at the C(34,35) diol. We also opted to transform the C(36) N-Boc moiety to the required N,N-dimethylamine prior to union with the C(9-25) spiroketal, based on preliminary experiments suggesting that 2-trimethylsilylethyl protection of the C(17) phosphate ester would not survive the Boc deprotection-reductive amination sequence.43b Two additional considerations came to the fore. Interchange of the acetonide and DEIPS groups could not be effected with lactam (+)-49 or (-)-E; this operation was therefore postponed until after establishing the amide linkage. Retaining the ester (40) Corey, E. J.; Nicolaou, K. C.; Balanson, R. D.; Machida, Y. Synthesis 1975, 590. (41) Flynn, D. L.; Zelle, R. E.; Grieco, P. A. J. Org. Chem. 1983, 48, 2424.

65 to furnish amide (+)-68 in 75% yield. Removal of the N-Boc group with TMSOTf,44 followed by reductive amination (HCHO, NaBH3CN, aqueous HOAc, CH3CN) introduced the two methyl groups at nitrogen with partial acetonide hydrolysis. Complete acetonide hydrolysis was readily achieved with HCl in anhydrous MeOH/ether. Silylation with diethylisopropylsilyl triflate (DEIPSOTf) in the presence of 2,6-lutidine then furnished (+)69 in 50% yield for the four steps, with only one chromatographic purification. Reduction of the ester was next achieved with LiAlH4 at -78 °C to afford alcohol (+)-70 (Scheme 21); low-temperature workup with ethyl acetate was essential to prevent loss of one of the DEIPS groups. Turning next to ylide formation, use of both the C(26) iodide or bromide (e.g., I2, PPh3 or CBr4, PPh3) was feasible, albeit they proved unstable. On the other hand the chloride obtained by treatment of (+)-70 with methanesulfonyl chloride in the presence of triethylamine and benzyltriethylammonium chloride was stable at 0 °C for weeks. Displacement with tri-n-butylphosphine in the mixed solvent system of CH3CN/THF (2:1) led to clean conversion to phosphonium salt (+)-DE in 95% yield. Wittig Coupling: A Model Study. Initially we explored the triphenyl- and tributylphosphonium salts 72a and 72b, generated in situ from iodide (+)-67 (Scheme 22). Of particular concern was the E/Z coupling selectivity and the need for C(17) hydroxyl protection. Reaction with hydrocinnamaldehyde revealed a dramatic increase in the E/Z ratio with the tri-n-butylphospho(42) The acetonide required strongly acidic conditions (CSA, MeOH, 45 °C, 24 h or 1.2 M aqueous HCl, THF, 50 °C, 12 h) to effect removal from model compound i, casting uncertainty upon its successful removal from fully protected calyculin. Indeed, although model compound iii could be deprotected, removal of the acetonide was unsuccessful with fully protected calyculin A. (b) Proton NMR comparison showed that cyclopentylidene hydrolysis from ii was 4.9 times faster than acetonide hydrolysis from i.

(43) (a) Toshima, K.; Mukaiyama, S.; Kinoshita, M.; Tatsuta, K. Tetrahedron Lett. 1989, 30, 6413. (b) Spoors, P. G.; Smith, A. B., III, unpublished results. (44) Hamada, Y.; Kato, S.; Shioiri, T. Tetrahedron Lett. 1985, 26, 3223.

Synthesis of (+)-Calyculin A and (-)-Calyculin B Scheme 22

J. Am. Chem. Soc., Vol. 121, No. 45, 1999 10485 Scheme 24

nium salt, consistent with the observations subsequently reported by Armstrong and co-workers.45,46 We also observed that the yield increased to 69% without significantly altering the E/Z selectivity when three equivalents of n-BuLi were employed. With the more advanced model spiroketals (+)-74 and (+)7547 (Scheme 23), moderate yields of the expected alkene Scheme 23 Introduction of the C(1-8) Cyanotetraene. To attach the C(3-8) subunit, the pivaloate group was removed from the C(9) terminus of (+)-BCDE via reduction (DIBAL or LiEt3BH) (Scheme 25). Initially this process proved capricious;51 reliable Scheme 25

products (+)-76 and (+)-77 were obtained with similar E selectivities. Although the experiment with (+)-74 established the compatibility of the Wittig coupling without protection at C(17),48 better yields were obtained with protection. Union of (+)-BC with (+)-DE. Aldehyde (+)-78, readily obtained in high yield from (+)-BC1 by removal of the C(25) O-benzyl protecting group and Ley oxidation [catalytic perruthenate (TPAP), NMO],49 was submitted to Wittig olefination (Scheme 24). As with the model studies, olefination with triphenylphosphonium salt 7950 (LiHMDS, THF) furnished (+)BCDE with 2:1 E/Z selectivity, whereas the tri-n-butylphosphonium salt (+)-DE provided a 5:1 E/Z mixture. With N,Ndimethylformamide as solvent, the E/Z ratio improved further to 9:1, consistent with the results of Armstrong et al.45 Evans and co-workers reported Wittig olefination with a related substrate (LDA, DMF, 0 °C) to be completely stereoselective. In our hands, their conditions afforded 10:1 E/Z selectivity, albeit the reaction was less efficient (∼32-69%). For material advancement we employed LiHMDS (1.5 eq) in DMF at 0 °C; isomer separation was achieved via radial chromatography. (45) Zhao, Z.; Scarlato, G. R.; Armstrong, R. W. Tetrahedron Lett. 1991, 32, 1609. (46) The role of the phosphorous ligands on the stereoselectivity in Wittig reactions has been discussed in detail, see: Vedejs, E.; Marth, C. F. J. Am. Chem. Soc. 1988, 110, 3948. (47) Compounds 74 and 75 were prepared from i, the latter prepared during the synthesis of the C(9-25) fragment (+)-BC (see ref 1). O-Methylation, hydrogenation (Pearlman’s catalyst) to saturate fully the alkyne, and removal of the C(25) O-benzyl afforded the C(25) primary alcohol. Ley oxidation (TPAP/NMO), with and without prior removal of the C(17) TBS group provided 74 and 75, respectively.

(48) Independently, Armstrong demonstrated similar transformations with the C(17) hydroxyl unprotected, see: Ogawa, A. K.; DeMattei, J. A.; Scarlato, G. R.; Tellew, J. E.; Chong, L. S.; Armstrong, R. W. J. Org. Chem. 1996, 61, 6153. (49) Ley, S. V.; Norman, J.; Griffith, W. P.; Marsden, S. P. Synthesis 1994, 639. (50) Phosphonium salt 79 was prepared from the corresponding alkyl chloride (+)-71 (PPh3, MeCN/THF; 28% yield).

results were eventually obtained by minimizing the amount of DIBAL followed by workup with sodium potassium tartrate. Ley oxidation49 then afforded aldehyde (+)-80 in high yield (84%). It should be emphasized that this oxidation proceeded in excellent yield despite the C(36) tertiary amine.52 Horner-Emmons coupling was then achieved via metalation of A with n-BuLi; a deep red-orange solution resulted. Addition of (+)-80 led to olefination within 30 min, with high selectivity (>10:1).53 With basic workup (aqueous NaHCO3), the product containing the C(3) carbonyl function masked as an enol ether could be isolated ( ∼50-60%) and chromatographed. However, best results entailed direct conversion via acid treatment (0.5 N HCl) to ketone (+)-81 (92% yield; E/Z ) 15:1). (51) This reductive deprotection initially provided variable yields and low mass balance. We speculate that chelation of the Lewis acidic boron and aluminum reagents by the C(36-37) permethylated 1,2-amino alcohol increased the product hydrophilicity and interfered with isolation; related C(9-25) substrates not possessing the 1,2-amino alcohol (i.e., before Wittig coupling) behaved normally. (52) If the C(36) tertiary amine is oxidized, the N-oxide product may be expected to be functionally similar to NMO. We did not detect the N-oxide. (53) Determination of the E/Z selectivity by 1H NMR spectroscopy was unreliable at this stage due to decomposition and the presence of interfering resonances.

10486 J. Am. Chem. Soc., Vol. 121, No. 45, 1999 Completion of the Calyculin Synthetic Venture. The stage was now set for Peterson olefination and global deprotection. In the event (Scheme 25), the desired cyanotetraene emerged in 94% yield as a 1.7:1 E/Z mixture. The Z and E isomers, corresponding to fully protected ent-calyculin A and entcalyculin B respectively, were easily separable by careful radial chromatography. All that remained was global deprotection. We recount here two important experiments which had earlier suggested the feasibility of this scenario. Oxazole (+)-69, when subjected to hydrolysis with HF (1.6 M in 8:1 CH3CN/H2O),54 effectively afforded 82 (Scheme 26). In a similar fashion, treatment of (+)Scheme 26

Smith et al. after 2 d, the major component present contained only one trimethylsilylethyl-protecting group.55 Prolonged exposure to HF (2 M in 8:1 CH3CN/H2O, 7 d) cleanly provided synthetic (+)-1 in 69% yield. In a similar fashion, fully protected ent-calyculin B afforded synthetic (-)-calyculin B (2) (84% yield). Synthetic (+)-calyculin A and (-)-calyculin B were identical in all respects (1H and 13C NMR, IR, UV, HRMS, TLC, HPLC), except for chiroptic properties,56 to natural (-)-1 and (+)-2, respectively.57 Acknowledgment. Support was provided by the National Institutes of Health (National Cancer Institute) through Grant CA-19033 and postdoctoral fellowships to G.K.F. and K.G.H. We also thank DuPont Pharmaceuticals, Inc. and Zeneca Pharmaceuticals, Inc. for support of this work. In addition, we thank Drs. George T. Furst and Patrick J. Carroll and Mr. John Dykins of the University of Pennsylvania Spectroscopic Service Center for assistance in securing and interpreting high-field NMR spectra, X-ray crystal structures, and mass spectra, respectively. Supporting Information Available: Spectroscopic and analytical data and selected experimental procedures for all synthetic intermediates (PDF). This material is available free of charge via the Internet at http://pubs.acs.org. JA992135E

BCDE with the same HF conditions led to 83 in 74% yield; prolonged exposure to HF was required to remove the remaining protecting group from the phosphate. The latter reaction established both the lability of the phosphate protection and the stability of the phosphate P-O(17) bond under the HF conditions. Bolstered by these results, fully protected ent-calyculin A (Scheme 25) was exposed to HF (1.5 M in 10:1 CH3CN/H2O); (54) We benefited significantly in this regard from the previous successes of Evans (see ref 9) and Masamune (see: Tanimoto, N.; Gerritz, S. W.; Sawabe, A.; Noda, T.; Filla, S. A.; Masamune, S. Angew. Chem., Int. Ed. Engl. 1994, 33, 673), each of whom demonstrated that calyculin A is stable to HF in aqueous acetonitrile.

(55) Interestingly, this monoprotected intermediate was more polar (TLC) than fully deprotected calyculin A. Together with precedent, this TLC behavior suggests that a fully deprotected phosphate is required for formation of a folded tertiary structure which exposes a less polar surface. Evans and coworkers observed that an intermediate with one t-butyldimethylsilyl group remaining coeluted with calyculin A on silica gel; it was proposed that this silyl group [tentatively assigned as C(11)-OTBS] was buried within a folded calyculin structure, minimizing the effect on the apparent overall polarity and accessibility to external reagents.9 In our case, the phosphate group is last to be deprotected. Deprotection at C(11) appears normal; this implicates the free phosphate as critical for the folding process. Similar behavior was observed during deprotections of calyculins A and B. (56) Synthetic (+)-calyculin A and (-)-calyculin B had optical rotations and CD spectra of equal magnitudes and opposite signs relative to the corresponding natural products. (57) We thank Professors Nobuhiro Fusetani and Shigeki Matsunaga (University of Tokyo) for authentic samples of natural (-)-1 and (+)-2, and for confirming that natural 2 has an [R] of +15° (c 0.07). Natural 2 was originally reported to have [R] -60° (c 0.05). We also thank Professor David Evans (Harvard University) for a sample of synthetic (+)-1.